I've been doing some analysis of character frequency, and on that basis have the following recommendations for changes to the exemplar characters.
As a reminder, the main exemplar characters are those that are required for native words in modern customary use. For example, "a-z" suffice for English. The aux exemplar characters include other characters that would not be unexpected in common non-technical publications: those that are native but not required (eg, ö as in coöperate, or "pronunced dāvĭs"), foreign loan words (eg, résumé). [A useful source of the aux letter are newspaper style guidelines.]
There is a breakdown on http://www.unicode.org/repos/cldr-tmp/trunk/dropbox/mark/exemplars/summary.html. Note the following:
Here are my suggestions. Please send feedback to firstname.lastname@example.org with any other suggestions, or add comments to http://unicode.org/cldr/trac/ticket/2789.
1. For all Latin-script languages, add [a-z] into the aux set. For others, check that any Latin script characters are deliberate, and either include all of a-z or none (I found in Tamil: <exemplarCharacters type="auxiliary">[a g i m t]</exemplarCharacters>)
2. zh shows the following as high-frequency characters but not in exemplars 网 机 产 册 没 只 帖 万 ... . Consider adding to main. There are some other high frequency characters in aux, that probably should be in main: 线 录 户
3. I collected some draft info on languages* we don't have, expressed in code at the end of this document. Consider adding locales to at least encompass this information.
4. ja doesn't include the following, should probably be in aux: 岡 阪 奈 藤 俺 伊 誰...
5. es should have ª in main, check that aux covers (French/German/Danish/Port.)
6. de should have ß in main, check aux covers French/Spanish/Danish/Turkish
7. pt should have ª º in main, French/Spanish/German/Danish in main
8. ko should have jamo in aux, and: 中 人 北 大 女 完 文 日 本 的 美 語...
9. it should check French/German/Danish in aux
10. In general, we should see which languages follow the convention of using trema to separate digraph vowels (eg naïve), and add the 6 vowels with trema to aux, at least.
11. in aux should cover French/Dutch
12. tr aux should cover French/German
13. zh-Hant: should look at 只 帖 搜 壇 ..
14. nl aux should cover French/Spanish/German/Danish
15. pl aux should cover French/German
16. fil aux should cover Spanish/French
17. qu* aux should cover Spanish
18. hu aux should cover German
19. el aux should cover polytonic greek
20. fi aux should cover French, German
21. We should include non-Western decimal digits into the corresponding exemplars
22. fa aux should include Arabic; ar aux should include Persian
23. da aux should cover French/German/Spanish
24. ca aux should cover Spanish
25. All Cyrillic aux should cover Russian
26. eu (Basque) aux should cover Spanish/French
27. ku-Arab aux should cover Arabic/Persian
28. br (Breton) aux should cover French
From Bug 1947, for reference.
The exemplar character set for ja appears to be too small.
1. It contains about 2,000 characters (Kanji, Hiragana and Katakana).
3. It does not contain <U+30F7, U+30FA> ('composed Katakana letters'), U+30FB
For instance, characters like U+4EDD,U+66D9, U+7DBE are not included although
While I was at it, I also looked at zh* and ko. All of them have about 2000
It's rather inconvenient to type hundreds (if not thousands) of characters in
Jungshik and I discussed this, and there are three possible sources (for each of Chinese (S+T), Japanese, and
1. charsets (in the case of Japanese, this would be probably: JIS 208 + 212 + 213. (This would be a large set, and
1a. Only use JIS 208. (The current approach appears to be JIS 208, but only level 1.)
2. Use the educational standards in each country/territory for primary+secondary requirements. We'd have to
3. Use the NIC restrictions for each country.
These would all overlap to a large degree, but wouldn't be the same. One possibility is to issue a PRI for public
There is a fourth possibility: Use the characters that are supported by the commonly-used fonts on various